A joint calibration method for runoff generation and runoff parameters in areas without measured flow rates
By combining regional rainfall and runoff experience with historical flood information, and employing multi-frequency calibration of runoff generation parameters and joint calibration of confluence parameters, the reliability problem of hydrological model parameter calibration in areas without measured flow rates was solved, achieving stable simulation accuracy and reliable input of flood processes under multi-frequency scenarios.
Patent Information
- Application Number
- CN202611569607.0
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-10-12
- Publication Date
- 2026-11-10
AI Technical Summary
In areas without measured flow rates, the calibration of hydrological model parameters in existing technologies lacks reliability. It suffers from strong subjectivity, large differences in parameter values, simulation results with different parameters having the same effect, difficulty in maintaining stable accuracy under multiple frequency scenarios, and failure to effectively utilize regional information for unified constraints.
By combining regional rainfall-runoff empirical relationships and historical flood information, a multi-frequency rate-determined runoff generation parameter is adopted, and the runoff parameters are determined in stages with peak flow and peak occurrence time as constraints, thus forming a multi-objective optimization method.
It improves the objectivity and repeatability of parameter determination, ensures the applicability and consistency of model parameters under different rainfall levels, and provides more reliable flood process input.
Smart Images

Figure CN123069719A_ABST
Abstract
Description
Technical Field
[0001] This invention belongs to the field of hydrological model parameter identification technology, and in particular relates to a method for joint calibration of runoff generation and confluence parameters in areas without measured flow rates. Background Technology
[0002] In hydrological modeling applications, hydrological modeling systems (such as HEC-HMS) are often used to construct runoff generation and confluence models to simulate watershed rainfall-runoff processes. Runoff generation calculations typically employ the SCS-CN curve method, while confluence calculations often use the SCS unit hydrograph method. For watersheds with hydrological stations and continuously measured rainfall-discharge hydrographs, model parameters are usually calibrated directly based on measured data, resulting in high parameter reliability. However, for areas lacking measured flow monitoring stations, the common practice is to estimate runoff generation parameters by consulting tables based on land use and soil types, and to estimate confluence parameters using empirical formulas based on watershed geographical characteristics (such as main channel length and watershed slope), or to manually adjust parameters based on a few historical floods. In engineering practice, when it is necessary to derive design flood processes at different design frequencies in areas without measured flow, these empirical parameters are often directly used in the hydrological model, and the model output is used as the upstream boundary input condition for subsequent hydrodynamic simulations or flood control analyses.
[0003] However, the existing technologies still have the following shortcomings when calibrating model parameters in areas without measured flow rates: First, due to the lack of continuous measured flow process lines as a calibration basis, the determination of runoff generation and runoff parameters heavily relies on empirical table lookups or manual calculations, leading to strong subjectivity in parameter values. Different technicians may obtain significantly different results, resulting in insufficient objectivity and repeatability. Second, runoff generation parameters mainly control the total runoff, while runoff parameters mainly control the temporal pattern of the flood process. There is a mutual compensation effect between the two. If they are determined separately, the deviation generated by one set of parameters may be compensated by the other set of parameters, resulting in the problem of "different parameters with the same effect" where different parameter values produce similar simulation output results, thus masking the physical distortion of the parameters themselves. Third, existing technologies usually only use a single design frequency or a single historical flood for parameter determination, making the calibrated parameters perform well only under that specific rainfall level. It is difficult to maintain stable simulation accuracy under multiple frequency scenarios such as frequent heavy rain, moderate heavy rain, and extreme heavy rain. Fourth, in the selection of targets for parameter calibration, only the fit of peak flow is often considered, while the constraint on the timing of peak occurrence is ignored. This results in simulated flood processes with similar peak amplitudes but significant deviations in peak occurrence time. For multi-tributary flood superposition analysis and joint gate operation, this significantly affects the judgment of the peak water level and the highest water level after superposition. Fifth, the abundant information existing in the region, such as rainfall-runoff empirical relationships, historical rainstorm records, flood trace survey data, and river cross-section measurement data, is currently only referenced in a scattered and qualitative manner. A unified constraint framework that can be calculated and quantitatively verified has not been formed to guide the joint calibration of model parameters. Summary of the Invention
[0004] This invention aims to address the lack of reliable calibration basis for HEC-HMS runoff generation and confluence parameters in areas without measured flow rates. Without relying on continuous measured flow process lines, it transforms regional statistical regularities and local historical flood response information into quantifiable constraints, forming a set of parameter joint calibration methods that can simultaneously constrain total runoff generation, peak amplitude, and flood timing, and can be stably applied at various design frequencies.
[0005] To address the aforementioned technical problems, this invention provides a method for joint calibration of runoff generation and confluence parameters in areas without measured flow rates, comprising: Based on the rainfall data of the design frequency and the empirical relationship between regional rainfall and runoff, the runoff generation parameters of the runoff generation model are calibrated at multiple frequencies to obtain the calibrated runoff generation parameters. Based on the calibrated runoff generation parameters, historical rainfall data, and reference values for peak flow and peak occurrence time derived from historical flood trace data, the runoff parameters of the runoff model are jointly calibrated using multiple objectives to obtain the calibrated runoff parameters. The calibrated runoff generation parameters and the calibrated confluence parameters are input into the hydrological model, and the design flood process results of the target inlet are output.
[0006] Preferably, the process of obtaining calibrated production flow parameters includes: Based on the rainfall data for the design frequency and the empirical relationship between rainfall and runoff in the region, the reference runoff depth for each design frequency is calculated. Based on the preset search range of proliferative flow parameters, candidate proliferative flow parameter values are generated; Input the candidate runoff generation parameter values and the design frequency of rainfall into the runoff generation model to obtain the simulated runoff depth of each candidate runoff generation parameter value at each design frequency. Based on the simulated runoff depth and the corresponding reference runoff depth at each design frequency for each candidate runoff parameter value, calculate the multi-frequency comprehensive error of each candidate runoff parameter value; Based on the multi-frequency integrated error, the calibrated flow parameters are selected from multiple candidate flow parameter values.
[0007] Preferably, the process of calculating the multi-frequency comprehensive error of each candidate runoff parameter value includes: Calculate the single-frequency error between the simulated runoff depth and the corresponding reference runoff depth for each candidate runoff parameter value at each design frequency; For each candidate flow generation parameter value, the single-frequency error under all design frequencies is summarized and calculated according to the preset frequency weight to obtain the multi-frequency comprehensive error.
[0008] Preferably, the process of obtaining the reference value of peak flood discharge derived from historical flood data includes: Based on the flood mark elevation and the corresponding river cross-section geometry, determine the cross-sectional area and hydraulic radius of the water passage. Based on the measurement data of the river section where the river cross-section is located, determine the water surface slope and the river roughness. Based on the cross-sectional area of the water passage, the hydraulic radius, the water surface slope, and the river roughness, the peak flow inversion value corresponding to the flood mark point is calculated.
[0009] Preferably, the process of obtaining the reference value of peak flood discharge derived from historical flood data further includes: For the same historical flood and multiple floodmark points within the same confluence area, the peak flow inversion values corresponding to each floodmark point are comprehensively summarized to obtain the peak flow reference value.
[0010] Preferably, the process of obtaining calibrated merge parameters includes: Calculate the initial values of the runoff parameters based on the geographical characteristics of the watershed and the calibrated runoff parameters; Based on the initial values of the merging parameters, a search space containing multiple candidate merging parameter combinations is constructed; Keeping the calibrated runoff parameters unchanged, for each candidate runoff parameter combination, the hydrological model is run based on the historical rainfall data to obtain the simulated peak flow and simulated peak time corresponding to each candidate runoff parameter combination; A comprehensive evaluation index is constructed based on the simulated peak flow and simulated peak time corresponding to each candidate confluence parameter combination, as well as the reference values of the peak flow and peak time. Based on the comprehensive evaluation index, the calibrated flow parameters are selected from multiple candidate flow parameter combinations.
[0011] Preferably, the process of calculating the initial values of the confluence parameters includes: The initial values of the runoff parameters are calculated based on the length of the main channel, the average slope of the basin, and the maximum water storage capacity of the basin obtained by converting the runoff generation parameters as calibrated.
[0012] Preferably, the process of constructing comprehensive evaluation indicators includes: Calculate the peak deviation between the simulated peak flow and the reference value of the peak flow, and the peak time deviation between the simulated peak time and the reference value of the peak time; Based on the preset flood peak deviation weight and peak time deviation weight, the flood peak deviation and the peak time deviation are weighted and summarized to obtain the comprehensive evaluation index.
[0013] Preferably, the method further includes: verifying the calibrated runoff generation parameters and the calibrated confluence parameters using historical flood data and regional design flood peak results.
[0014] Preferably, after outputting the design flood process results of the target inlet, the method further includes: The design flood process results are converted into the boundary condition data format required by the hydrodynamic simulation software.
[0015] Compared with the prior art, the present invention has the following advantages and technical effects: This invention uses the empirical relationship between regional rainfall and runoff as an independent quantitative constraint on the total runoff, and uses the peak flow inverted from historical flood marks and the peak occurrence time inferred from surveys as independent quantitative constraints on the confluence process. This makes the model parameters no longer dependent on purely empirical values or manual calculations when there is no measured flow, which significantly improves the objectivity and repeatability of parameter determination.
[0016] This invention performs phased calibration according to the technical dependence sequence of "first generating flow, then merging flow". After the generating flow parameters are fixed, the merging flow parameters are jointly optimized. This effectively cuts off the mutual compensation path between generating flow error and merging flow error, and reduces the interference of different parameters with the same effect on the physical authenticity of the parameters.
[0017] In the process of calibrating runoff parameters, this invention incorporates rainfall data from multiple different design frequencies (covering common extreme rainstorm scenarios) and uses the minimum comprehensive error of multiple frequencies as the selection criterion to ensure that the calibrated runoff parameters have good applicability and consistency under different rainfall levels.
[0018] In the process of calibrating confluence parameters, this invention incorporates both peak flow error and peak occurrence time error into the same comprehensive objective function to take into account both the amplitude accuracy and timing accuracy of the flood process, thereby providing a more reliable flood process input for multi-tributary flood superposition analysis and joint gate scheduling.
[0019] This invention transforms multi-source data, such as regional empirical relationships, design rainfall, historical rainfall, flood trace surveys, and cross-sectional measurements, into calculable model constraints. These constraints are then used in parameter search and error calculation according to clear mathematical rules, forming a complete closed-loop framework from data input and parameter calibration to result verification. This allows for the full and traceable utilization of various types of field information in engineering projects. Attached Figure Description
[0020] The accompanying drawings, which form part of this application, are used to provide a further understanding of this application. The illustrative embodiments and descriptions of this application are used to explain this application and do not constitute an undue limitation of this application. In the drawings: Figure 1 This is a schematic diagram of the method flow according to an embodiment of the present invention; Figure 2 This is a comparison chart of simulated runoff depth and regional empirical values for different CN values in a small canal according to an embodiment of the present invention; Figure 3 This is a diagram showing the comprehensive error curve and optimal value of the CN value for the small canal according to an embodiment of the present invention. Figure 4 This is a comparison chart of simulated runoff depth and regional empirical values for the Zhaowang River under different CN values according to an embodiment of the present invention. Figure 5 This is a diagram showing the comprehensive error curve and optimal value of the CN value of Zhao Wanghe River according to an embodiment of the present invention. Figure 6 This is a comparison chart of simulated runoff depth and regional empirical values for Zhougong River under different CN values according to an embodiment of the present invention; Figure 7 This is a comprehensive error curve and optimal value diagram of the CN value of Zhougong River in an embodiment of the present invention. Detailed Implementation
[0021] It should be noted that, unless otherwise specified, the embodiments and features described in this application can be combined with each other. This application will now be described in detail with reference to the accompanying drawings and embodiments.
[0022] It should be noted that the steps shown in the flowchart in the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions, and although a logical order is shown in the flowchart, in some cases the steps shown or described may be executed in a different order than that shown here.
[0023] like Figure 1 As shown, this embodiment provides a method for joint calibration of runoff generation and confluence parameters in areas without measured flow rates, including: Based on the rainfall data of the design frequency and the empirical relationship between regional rainfall and runoff, the runoff generation parameters of the runoff generation model are calibrated at multiple frequencies to obtain the calibrated runoff generation parameters. Based on the calibrated runoff generation parameters, historical rainfall data, and reference values for peak flow and peak time derived from historical flood trace data, the runoff parameters of the runoff model are jointly calibrated using multiple objectives to obtain the calibrated runoff parameters. Input the calibrated runoff generation parameters and calibrated runoff parameters into the hydrological model, and output the design flood process results at the target inlet.
[0024] Furthermore, this embodiment specifically relates to a method for jointly calibrating the SCS-CN runoff generation parameters and SCS unit hydrograph watershed lag parameters in the HEC-HMS model in areas lacking continuous measured flow process lines, utilizing regional rainfall-runoff empirical relationships, design storm data, historical storm data, historical flood marks, and river cross-section information, in stages, at multiple frequencies, and with multiple objectives, and outputting the design flood process at the confluence of level rivers. Specifically, it transforms model calibration under the condition of "no measured flow" into a problem of "staged joint optimization under multi-source indirect constraints." First, it uses regional rainfall-runoff empirical curves to provide total runoff generation constraints, searching for CN values at multiple design frequencies; after fixing the calibrated CN values, it uses the results of inverting peak flow and peak occurrence time from historical flood marks to jointly search for watershed lag; finally, it uses historical event errors and multi-design frequency errors to double-verify the parameter combination.
[0025] Furthermore, the process of obtaining calibrated flow parameters includes: Based on the rainfall data for the design frequency and the empirical relationship between regional rainfall and runoff, the reference runoff depth for each design frequency is calculated. Based on the preset search range of proliferative flow parameters, candidate proliferative flow parameter values are generated; Input the candidate runoff generation parameter values and the design frequency of rainfall into the runoff generation model to obtain the simulated runoff depth for each candidate runoff generation parameter value at each design frequency. Based on the simulated runoff depth and the corresponding reference runoff depth at each design frequency for each candidate runoff parameter value, calculate the multi-frequency comprehensive error of each candidate runoff parameter value; Based on the multi-frequency integrated error, the calibrated production flow parameter is selected from multiple candidate production flow parameter values.
[0026] Furthermore, the input data in this embodiment includes at least basic watershed and model data, design storm data, historical storm data, regional rainfall-runoff empirical relationship data, historical flood marks and river cross-section data, and calibration control parameters. All types of data are processed according to a unified sub-watershed number, spatial coordinate reference, and time reference before being input into the model. Specifically, this includes: (1) Basic data of the basin and model: the boundary, area, topographic elevation, river network structure, main channel length, longitudinal slope, confluence location and upstream and downstream connection relationship of the basin or sub-basin to be calibrated; based on the above data, establish sub-basins, river segments, confluence nodes and outlet nodes in the HEC-HMS model, and configure the SCS-CN runoff generation module and SCS unit line confluence module.
[0027] (2) Design rainfall data: Design surface rainfall and design rainfall pattern corresponding to multiple design frequencies, including total rainfall, time history distribution and calculation time step under design frequencies such as 20%, 10%, 5%, 2% and 1%; when the original rainfall pattern time step is inconsistent with the model step step, the design rainfall pattern is converted to time discretization or interpolation with equal total amount.
[0028] (3) Historical rainstorm data: hourly or shorter time scale rainfall sequences, rain gauge coordinates, data integrity records and typical historical rainstorm events of the study area and adjacent rain gauge stations; based on the Thiessen polygon method, isohyetal method or other areal rainfall calculation methods, the historical areal rainfall process of each sub-basin is generated, and the preceding impact rainfall of each event is recorded.
[0029] (4) Regional empirical relationship data: applicable to the rainfall-runoff empirical curves, design flood calculation results or other regionally verified runoff depth estimation relationships in the study area, as well as the corresponding applicable scope, frequency conditions and interpolation rules; this type of data is used to calculate the reference runoff depth for each sub-basin and each design frequency, as an indirect constraint for the calibration of runoff generation parameters.
[0030] (5) Flood marks and river cross-section data: the spatial location of historical flood mark points, flood mark elevation, flood mark type, preservation status and measurement date, as well as the cross-sectional coordinates, water flow area, wetted perimeter, hydraulic radius, riverbed and bank slope composition, water surface gradient and Manning roughness range of the corresponding cross-section; and also including resident interviews, water conservancy department records and data that can help determine the time of flood occurrence and peak time.
[0031] (6) Calibration control parameters: candidate search range, search step size, and weights of multiple design frequencies for runoff generation parameters; empirical initial value, candidate search range, and search step size for watershed lag time; and calculation methods and weight configurations for runoff depth error, peak flow error, and peak occurrence time error. In this embodiment, the runoff generation parameter search range can be set to 30–98 with a step size of 1, and the watershed lag time candidate range can be set to 0.5–2.0 times the empirical initial value. However, the above values can be adjusted according to watershed characteristics and calculation accuracy requirements.
[0032] (7) Verification data: reference values of peak flow, peak time, and regional design peak results corresponding to historical events are used to verify the model after calibration using historical events and multiple design frequencies. During the verification, the original input, simulation output, error details, and parameter combinations of the model are saved simultaneously to ensure that the calibration process is traceable and reproducible.
[0033] Furthermore, the process of calculating the multi-frequency integrated error for each candidate runoff parameter value includes: Calculate the single-frequency error between the simulated runoff depth and the corresponding reference runoff depth for each candidate runoff parameter value at each design frequency; For each candidate flow generation parameter value, the single-frequency error under all design frequencies is summarized and calculated according to the preset frequency weight to obtain the multi-frequency comprehensive error.
[0034] Furthermore, the process of obtaining reference values for peak flood discharge derived from historical flood data includes: Based on the flood mark elevation and the corresponding river cross-section geometry, determine the cross-sectional area and hydraulic radius of the water passage. Based on the measurement data of the river section where the river cross-section is located, determine the water surface slope and the river roughness. The peak flow inversion value corresponding to the flood mark point is calculated based on the cross-sectional area of the water passage, the hydraulic radius, the water surface slope, and the river roughness.
[0035] Furthermore, the process of obtaining reference values for peak flood discharge derived from historical flood data also includes: For the same historical flood and multiple floodmark points within the same confluence area, the peak flow inversion values corresponding to each floodmark point are comprehensively summarized to obtain a reference value for the peak flow.
[0036] Furthermore, the process of obtaining the calibrated confluence parameters includes: Calculate the initial values of the runoff parameters based on the geographical characteristics of the watershed and the calibrated runoff generation parameters; Based on the initial values of the flow parameters, a search space containing multiple candidate flow parameter combinations is constructed; Keeping the calibrated runoff parameters unchanged, for each candidate runoff parameter combination, the hydrological model is run based on historical rainfall data to obtain the simulated peak flow and simulated peak time corresponding to each candidate runoff parameter combination; A comprehensive evaluation index is constructed based on the simulated peak flow and simulated peak time corresponding to each candidate confluence parameter combination, as well as the reference values of peak flow and peak time. Based on comprehensive evaluation indicators, the calibrated flow parameters are selected from multiple candidate flow parameter combinations.
[0037] Furthermore, the process of calculating the initial values of the confluence parameters includes: The initial values of the runoff parameters are calculated based on the length of the main channel, the average slope of the basin, and the maximum water storage capacity of the basin obtained by converting the runoff generation parameters from the calibrated parameters.
[0038] Furthermore, the process of constructing comprehensive evaluation indicators includes: Calculate the peak flow deviation between the simulated peak flow and the reference peak flow, and the peak time deviation between the simulated peak time and the reference peak time. Based on the preset peak deviation weight and peak time deviation weight, the peak deviation and peak time deviation are weighted and summarized to obtain a comprehensive evaluation index.
[0039] Furthermore, the method also includes: using historical flood data and regional design flood peak results to verify the calibrated runoff generation parameters and calibrated runoff parameters.
[0040] Furthermore, after outputting the design flood process results for the target inlet, it also includes: Convert the design flood process results into the boundary condition data format required by the hydrodynamic simulation software.
[0041] As an optional implementation, the method of this embodiment specifically includes the following steps: S101: Establish the HEC-HMS runoff generation and confluence model for the watershed with the stator, take the design rainstorm or historical rainstorm process as input, select the SCS-CN curve method as the runoff generation calculation method, select the SCS unit hydrograph method as the confluence calculation method, and the model output is the flood discharge process at the specified inlet.
[0042] S102: For multiple design frequencies, calculate the reference runoff depth for each sub-basin based on the relationship curves between design areal rainfall, anterior impact rainfall, and regional rainfall-runoff. Convert the design rainfall pattern into a time-series rainfall input that meets the model calculation step size requirements.
[0043] S103: Generate candidate CN values within a preset CN search interval at a given step size. Run SCS-CN calculation for each candidate CN value and each design frequency to obtain the cumulative simulated runoff depth, and calculate the error between the simulated runoff depth and the reference runoff depth.
[0044] S104: Summarize the runoff depth errors for each design frequency to construct a multi-frequency integrated objective function; select the candidate CN value with the smallest integrated objective function as the runoff generation parameter for the corresponding sub-basin. If the local optimum value for each frequency fluctuates slightly, determine the representative CN value based on the integrated objective function and the average trend.
[0045] S105: Select historical rainstorm events and generate hourly areal rainfall processes for each sub-basin using rain gauge data and areal rainfall calculation methods such as the Thiessen polygon method; calculate the initial value of basin lag time based on the main channel length, basin slope, and the maximum potential water storage converted from the CN value.
[0046] S106: For historical flood marks, determine the cross-sectional area, hydraulic radius, and water surface gradient based on the flood mark elevation and the corresponding river section, and calculate the peak flow corresponding to the flood mark level using the Manning formula; when there are multiple flood marks in the same flood and the same sub-basin, summarize the inversion results of each section to obtain a reference value for the peak flow; and infer the reference value for the peak time by combining the visit records and rainfall process.
[0047] S107: Keep the CN value determined in S104 unchanged, generate candidate watershed time combinations within a preset multiple range of the initial time delay value; run HEC-HMS for each combination and each historical rainstorm event, compare the simulated peak flow with the flood mark inverted peak flow, and the simulated peak time with the reference peak time, construct a comprehensive error function that takes into account both the peak amplitude and the flood timing, and select the time combination with the smallest comprehensive error as the calibration result.
[0048] S108: Input the CN value and watershed lag time obtained from calibration into HEC-HMS, run the model for multiple design frequencies, and output the flood flow process at the confluence of each level river; verify the design flood peak error using historical field errors and regional empirical formulas, and convert the verified flood process into the upstream boundary conditions of the hydrodynamic model.
[0049] Furthermore, referring to Figures 1 to 6 The calibration results shown, specifically the multi-frequency calibration process for the flow generation parameter CN in steps S103 and S104, includes: S1031: For each undetermined sub-basin, select multiple design frequencies, read the design areal rainfall, anterior impact rainfall and design rainfall pattern corresponding to each design frequency, and calculate the reference runoff depth according to the regional rainfall-runoff empirical relationship. S1032: Generate candidate CN values within the preset CN search interval according to a given step size, and combine each candidate CN value with the time-series rainfall input of each design frequency to form the model working condition to be run; S1033: Run each model condition sequentially to obtain the cumulative simulated runoff depth of each candidate CN value at each design frequency, and establish a corresponding relationship between sub-basins, candidate CN values and design frequencies; S1034: Compare the cumulative simulated runoff depth with the reference runoff depth at the same design frequency, and calculate the single-frequency error corresponding to each candidate CN value; wherein, the single-frequency error can be absolute error, relative error or weighted square error; S1041: Summarize the single-frequency error of the same candidate CN value at all design frequencies according to the preset frequency weight, and obtain the multi-frequency comprehensive objective function value of the candidate CN value; S1042: Compare the multi-frequency integrated objective function values of all candidate CN values, and determine the candidate CN value with the smallest integrated objective function value as the calibration CN value of the corresponding sub-basin; when the local optimum value fluctuates under different design frequencies, the representative CN value is determined by the integrated objective function and its overall trend.
[0050] For the Sub-basins and the first A design frequency, making H sim (k,i,CN) represents the cumulative runoff depth calculated using candidate CN values, H ref (k,i) represents the reference runoff depth obtained from the empirical relationship between regional rainfall and runoff. A single-frequency error can be constructed using absolute error, relative error, or weighted squared error, and then summed over all design frequencies to form a comprehensive target. The relevant formulas are as follows: (1) In the formula, for A multi-frequency composite objective function for the value; For designing frequency indexes; Weights for each frequency; For error measurement function, In the first i At a design frequency, the cumulative runoff depth is calculated using a candidate CN value and a runoff generation model. This is the reference runoff depth determined based on the empirical relationship between regional rainfall and runoff at the same design frequency.
[0051] Specifically, in this embodiment, the CN search range is 30–98, the search step size is 1, and the design frequencies are 20%, 10%, 5%, 2%, and 1%. The calibration CN values for the Xiaoyunhe River, Zhaowanghe River, and Zhougonghe River are 62, 63, and 66, respectively.
[0052] like Figure 2 As shown, taking the Xiaoyunhe sub-basin as an example, the cumulative simulated runoff depth calculated by different candidate CN values at five design frequencies of 20%, 10%, 5%, 2% and 1% is compared with the reference runoff depth determined by the regional rainfall-runoff empirical relationship; the closer the candidate CN value is to the reasonable range, the higher the degree of agreement between the simulated runoff depth curve and the reference runoff depth curve.
[0053] like Figure 3 As shown, the runoff depth errors of each candidate CN value of the Xiaoyun River at five design frequencies are summarized to obtain a multi-frequency comprehensive error curve; the lowest point of the curve corresponds to a CN value of 62, so CN=62 is determined as the calibration runoff generation parameter of the Xiaoyun River sub-basin.
[0054] like Figure 4 As shown, the cumulative simulated runoff depth and reference runoff depth under different candidate CN values in the Zhaowanghezi watershed are compared with each design frequency to obtain the runoff response and error of each candidate CN value under different rainfall levels.
[0055] like Figure 5 As shown, error curves were plotted based on the multi-frequency integrated errors corresponding to each candidate CN value of the Zhao Wang River. The lowest point of the curve corresponds to a CN value of 63. Therefore, CN=63 was determined as the calibration runoff parameter for the Zhao Wang River sub-basin.
[0056] like Figure 6 As shown, the cumulative simulated runoff depth and reference runoff depth under different candidate CN values in the Zhougonghezi watershed are compared at each design frequency to verify the applicability of the candidate CN values under various design rainstorm conditions.
[0057] like Figure 7 As shown in the figure, error curves were plotted based on the multi-frequency integrated errors corresponding to each candidate CN value of the Zhougong River. The lowest point of the curve corresponds to a CN value of 66. Therefore, CN=66 was determined as the calibration runoff parameter for the Zhougong River sub-basin. The results shown in the above figures correspond to the candidate parameter generation, model running, error calculation, comprehensive evaluation, and optimal parameter determination processes in steps S1031 to S1042.
[0058] Furthermore, flood mark inversion converts the observable flood mark water level into a reference value for peak flow that can be used for model calibration. For each flood mark section, the cross-sectional area is determined based on the flood mark elevation. The hydraulic radius R is obtained from the wetted perimeter; the water surface gradient is determined by combining the longitudinal profile of the river section or adjacent water level information. The Manning roughness coefficient n is selected based on the riverbed composition and cross-sectional characteristics. (2) In the formula, Let n be the inverted peak flow rate corresponding to the flood mark water level, n be the Manning roughness coefficient, and A be the cross-sectional area of the water passage. For hydraulic radius, The water surface gradient is used. For the same flood and multiple flood mark points within the same sub-basin, the average value of the peak discharge at each cross-section can be taken, or a weighted average value can be used after verifying the reliability of the flood mark.
[0059] To reduce the biases introduced by the reliability of flood marks, cross-sectional generalization, and roughness selection, it is preferable to classify the reliability of flood mark points, record the flood mark type, preservation status, measurement date, spatial correspondence between cross-section and flood mark, and perform sensitivity analysis on the value range of n and J.
[0060] Furthermore, after fixing the CN value, a candidate search interval is established centered on the initial lag value obtained from the empirical formula. The embodiment described in the document sets the search range to 0.5 to 2.0 times the initial value, and runs HEC-HMS on candidate lag combinations for the three sub-basins. Each combination outputs the simulated peak flow and peak occurrence time for each historical event. (3) In the formula, The watershed time-delay comprehensive objective function is... For a combination of time delay parameters for one or more sub-basins, To simulate the error of peak flood discharge relative to the peak flood discharge inverted from the flood trail, To simulate the error between the peak occurrence time and the reference peak occurrence time, and This is the weighting coefficient. Let J be... T The smallest parameter combination is taken as the calibration result. The original weights, error normalization method, and search step size should be determined according to the calculation procedure.
[0061] This embodiment avoids sacrificing the timing of the flood process by simultaneously using two objectives: peak amplitude and peak occurrence time.
[0062] After further calibration, the system will output at least the following: CN values for each sub-basin, basin lag time, historical peak flow error, peak occurrence time error, flood flow process at each design frequency inlet, corresponding peak flow and peak occurrence time. The flood process can be derived as a time-flow sequence and used as the upstream boundary of MIKE11 or other one-dimensional / two-dimensional hydrodynamic models.
[0063] When used for automated software implementation, data preprocessing, candidate parameter generation, HEC-HMS batch calling, result parsing, objective function calculation, optimal parameter selection, error verification, and boundary file export can be encapsulated into mutually callable functional modules; the modules exchange data through structured files, database records, or model interfaces.
[0064] In the absence of a continuous measured flow process line, this embodiment utilizes the empirical relationship between regional rainfall and runoff and historical flood marks to construct independent constraints, thereby reducing the uncertainty of parameters relying entirely on empirical values or manual calculations.
[0065] This embodiment connects the multi-frequency calibration of CN value with the joint calibration of peak-peak time of watershed lag according to the technical dependence order, so that the total runoff, peak amplitude and flood timing are constrained respectively, reducing the parameter compensation effect.
[0066] This embodiment evaluates the CN value across multiple design frequencies, enhancing the consistency of parameters across rainfall levels and avoiding applicability to a single design frequency.
[0067] This embodiment transforms field data such as flood marks, cross sections, and access records into flood peaks and time-series reference quantities that the model can calculate, thereby improving the engineering interpretability and reproducibility of model calibration in areas with limited data.
[0068] The output results of this embodiment can be directly used for the inlet flood process in the hydrodynamic simulation of the main canal, providing boundary conditions for the identification of local over-limit sections, flood contribution analysis, and optimization of gate control schemes.
[0069] In the above implementation method, as an additional implementation method, the research object is three sub-basins of level rivers along the Lubei Main Canal: the Xiaoyun River, the Zhaowang River, and the Zhougong River. All three sub-basins are located in the Lubei Plain, with relatively gentle terrain slopes. While measured flow data are insufficient, regional rainfall data, design flood calculation data, historical flood marks, and river cross-section data are available. A sub-basin model is constructed using HEC-HMS, with design rainfall or historical rainfall as input and the flood process at the confluence of the level rivers as output.
[0070] Design Rainfall and Reference Runoff Depth: Five design frequencies (20%, 10%, 5%, 2%, and 1%) were selected, corresponding to various engineering conditions ranging from routine flood control to extreme rainstorms. The design rainfall pattern was converted to a different time step, with other time steps broken down into 1-hour steps. Based on previous rainfall impacts, reference runoff depths for each sub-basin and design frequency were calculated using linear interpolation of empirical rainfall-runoff relationship curves applicable to the study area.
[0071] CN value calibration results: The CN value was traversed within the range of 30 to 98 with a step size of 1. Each candidate CN value was tested at five design frequencies, and the cumulative runoff depth and its comprehensive error relative to the reference runoff depth were calculated. Ultimately, the CN values for the Xiaoyun River, Zhaowang River, and Zhougong River were determined to be 62, 63, and 66, respectively. Although the local optimum values fluctuated slightly at different frequencies, the overall optimum value remained stable, indicating that the selected CN values can maintain a consistent runoff response under multiple rainfall levels.
[0072] Historical Rainfall, Flood Mark Inversion, and Lag Time Calibration: Hourly rainfall data from surrounding rain gauge stations from 1985 to 2015 were used to select eight typical historical rainstorms. The Thiessen polygon method was employed to calculate the isal rainfall process for each event and sub-basin. Field data included 19 historical flood mark points, corresponding river cross-sections, as well as data from resident interviews and water resources department records. Peak discharge was inverted using the Manning formula based on flood mark elevation, cross-sectional area, hydraulic radius, roughness coefficient, and water surface gradient; peak time was inferred by combining interview records and rainfall process data.
[0073] The initial time delays for the Xiaoyun River, Zhaowang River, and Zhougong River were 13.4 h, 35.8 h, and 7.9 h, respectively. The search range was 0.5 to 2.0 times the initial values to optimize the combination of the three time delay parameters. Keeping the CN value constant, and minimizing the combined objective of peak flow error and peak occurrence time error, the recommended time delays for the three sub-basins were 1360 min, 1970 min, and 1450 min, respectively.
[0074] Historical data verification: According to the data, the above errors meet the evaluation requirements of the adopted standard that the peak flow error does not exceed 20% and the peak occurrence time error does not exceed ±3h.
[0075] Verification and Engineering Output at Multiple Design Frequencies: The calibrated CN values and watershed delays were input into HEC-HMS to deduce the inlet flood processes under five design frequencies and compare them with the design peak discharges of regional empirical formulas. The peak discharge error ranges for the Xiaoyun River, Zhaowang River, and Zhougong River were 1.3%–9.3%, 0.8%–7.7%, and 0.4%–4.8%, respectively, all controlled within 10%.
[0076] Further results indicate that the peak flow of the Zhao Wang River is approximately 2.09 times that of the Xiaoyun River and 3.47 times that of the Zhougong River, with the ratio being less affected by changes in design frequency. The calibrated flood process can serve as the upstream boundary for hydrodynamic simulation of the main canal, used to analyze the superposition of floods from multiple inlets, local over-limit sections, and gate combination regulation.
[0077] In the above implementation, as an alternative implementation, the regional empirical relationship can be replaced by a regionally validated rainfall-runoff statistical model, a neighboring similar watershed transfer relationship, or other regionalized methods that can provide a reference runoff depth.
[0078] In the above implementation, as an alternative implementation, the areal rainfall can be calculated using methods such as Thiessen polygons, isohyetal lines, inverse distance ratio, and radar-rain gauge fusion; the alternative method should retain the temporal sequence of the rainstorm and the spatial representativeness of the sub-basin.
[0079] In the above implementation, as an alternative implementation, in addition to the Manning formula, the peak flow inversion can adopt one-dimensional or two-dimensional hydrodynamic inversion supported by cross sections and boundary conditions; multiple flood trace points can be estimated by averaging, weighted averaging or robust estimation based on credibility.
[0080] In the above implementation, as an alternative implementation, parameter search can employ traversal, grid search, genetic algorithm, particle swarm optimization, Bayesian optimization, or other deterministic / heuristic optimization methods, as long as the multi-frequency flow generation constraint and the joint constraint of peak-peak time are maintained.
[0081] In the above implementation, as an alternative implementation, the error function can be evaluated using absolute error, relative error, squared error, normalized mean square error, or multi-objective Pareto; the weights can be equally weighted or configured according to engineering risk.
[0082] In the above implementation, as an alternative implementation, the flow generation model and the flow confluence model can be replaced with model components that have equivalent input-processing-output relationships. However, the replacement scheme must be able to express the total flow generation and flow confluence timing respectively, and accept the corresponding multi-source constraints.
[0083] In the above implementation method, as an alternative implementation method, the project object is not limited to the Lubei Main Canal, but can be extended to other plain areas or small and medium-sized watersheds that lack flow stations but have regional experience relationships, historical rainstorms, flood marks and cross-sectional data.
[0084] The above are merely preferred embodiments of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.
Claims
1. A method for joint calibration of runoff generation and confluence parameters in areas without measured flow rates, characterized in that, include: Based on the rainfall data of the design frequency and the empirical relationship between regional rainfall and runoff, the runoff parameters of the runoff generation model are calibrated at multiple frequencies to obtain the calibrated runoff parameters. Based on the calibrated runoff generation parameters, historical rainfall data, and reference values for peak flow and peak occurrence time derived from historical flood trace data, the runoff parameters of the runoff model are jointly calibrated using multiple objectives to obtain the calibrated runoff parameters. The calibrated runoff generation parameters and the calibrated confluence parameters are input into the hydrological model, and the design flood process results of the target inlet are output.
2. The method according to claim 1, characterized in that, The process of obtaining calibrated production parameters includes: Based on the rainfall data for the design frequency and the empirical relationship between rainfall and runoff in the region, the reference runoff depth for each design frequency is calculated. Based on the preset search range of proliferative flow parameters, candidate proliferative flow parameter values are generated; Input the candidate runoff generation parameter values and the design frequency of rainfall into the runoff generation model to obtain the simulated runoff depth of each candidate runoff generation parameter value at each design frequency. Based on the simulated runoff depth and the corresponding reference runoff depth at each design frequency for each candidate runoff parameter value, calculate the multi-frequency comprehensive error of each candidate runoff parameter value. Based on the multi-frequency integrated error, the calibrated flow parameters are selected from multiple candidate flow parameter values.
3. The method according to claim 2, characterized in that, The process of calculating the multi-frequency integrated error for each candidate runoff parameter value includes: Calculate the single-frequency error between the simulated runoff depth and the corresponding reference runoff depth for each candidate runoff parameter value at each design frequency; For each candidate flow generation parameter value, the single-frequency error under all design frequencies is summarized and calculated according to the preset frequency weight to obtain the multi-frequency comprehensive error.
4. The method according to claim 1, characterized in that, The process of obtaining reference values for peak flood discharge derived from historical flood data includes: Based on the flood mark elevation and the corresponding river cross-section geometry, determine the cross-sectional area and hydraulic radius of the water passage. Based on the measurement data of the river section where the river cross-section is located, determine the water surface slope and the river roughness. Based on the cross-sectional area of the water passage, the hydraulic radius, the water surface slope, and the river roughness, the peak flow inversion value corresponding to the flood mark point is calculated.
5. The method according to claim 1, characterized in that, The process of obtaining reference values for peak flood discharge derived from historical flood data also includes: For the same historical flood and multiple floodmark points within the same confluence area, the peak flow inversion values corresponding to each floodmark point are comprehensively summarized to obtain the peak flow reference value.
6. The method according to claim 1, characterized in that, The process of obtaining calibrated bus parameters includes: Calculate the initial values of the runoff parameters based on the geographical characteristics of the watershed and the calibrated runoff parameters; Based on the initial values of the merging parameters, a search space containing multiple candidate merging parameter combinations is constructed; Keeping the calibrated runoff parameters unchanged, for each candidate runoff parameter combination, the hydrological model is run based on the historical rainfall data to obtain the simulated peak flow and simulated peak time corresponding to each candidate runoff parameter combination; A comprehensive evaluation index is constructed based on the simulated peak flow and simulated peak time corresponding to each candidate confluence parameter combination, as well as the reference values of the peak flow and peak time. Based on the comprehensive evaluation index, the calibrated flow parameters are selected from multiple candidate flow parameter combinations.
7. The method according to claim 6, characterized in that, The process of calculating the initial values of the bus parameters includes: The initial values of the runoff parameters are calculated based on the length of the main channel, the average slope of the basin, and the maximum water storage capacity of the basin obtained by converting the runoff generation parameters as calibrated.
8. The method according to claim 6, characterized in that, The process of constructing comprehensive evaluation indicators includes: Calculate the peak deviation between the simulated peak flow and the reference value of the peak flow, and the peak time deviation between the simulated peak time and the reference value of the peak time; Based on the preset flood peak deviation weight and peak time deviation weight, the flood peak deviation and the peak time deviation are weighted and summarized to obtain the comprehensive evaluation index.
9. The method according to claim 1, characterized in that, The method further includes: verifying the calibrated runoff generation parameters and the calibrated confluence parameters using historical flood data and regional design flood peak results.
10. The method according to claim 1, characterized in that, After outputting the design flood process results for the target inlet, the following is also included: The design flood process results are converted into the boundary condition data format required by the hydrodynamic simulation software.
